On this page
Intake fields are the questions your bot finds out from every new lead. Each one becomes a piece of structured data on the lead — which is what makes qualification rules and matching possible at all.
Edit them at Business setup → What your bot finds out, which Quick start links to under Fine-tune. The page groups your questions, shows why you ask each one and lets you steer how your bot asks it; that side is in What your bot finds out. This article is about the questions themselves.
What a question has
| Setting | What it does |
|---|---|
| Question | The human name. What you see; roughly what the bot asks |
| Type | What kind of answer is acceptable: short text, paragraph, number, currency, email address, phone number, links, single choice or multiple choice. See Field and rule types |
| Must know before your team calls | Whether the conversation can finish without it |
| Why you ask | The reason you ask, in a sentence. Your bot words the question from it |
| Category group | For choice fields, which list of options applies |
Everything else is under More on the question's row: what to do if the lead is not sure, how much it matters, when it applies, whether it is sensitive and, for an amount, what it covers. The page itself is explained in What your bot finds out.
The older AI prompt hint is still in the edit form. It is used only when Why you ask is empty and the question is not a core one. Write a reason instead.
"Must know" is a promise the conversation has to keep
A must-know question is one the bot keeps asking about until it has an answer, but not for ever: asked twice with no answer, it rests while the bot asks the others, comes back once at the end, and is then let go. A lead who says they have not decided yet ("we haven't decided on a region") is not pressed either, and a lead who is simply unsure is helped first. A shrug ("whatever") never lets the bot drop it. Marking every question must-know feels thorough and still produces long conversations. The lead plainly refusing ("I'd rather not share a budget") stops it at once: the bot accepts that, does not ask again, and holds the lead for your team with the question on their card under Not answered, so you can decide. A question let go without an answer is held the same way, under the same label. A lead is never turned away for an answer they did not give.
The test is: would you refuse this lead if you never learned this? If not, make it optional, and your team can ask later. An optional question is still asked of every lead, but only once or twice, and a lead who would rather not answer, or only shrugs, is let go.
Order is a preference
Your bot works from what it knows, not from the order on the page. If a lead answers question four inside their opening message, question four is done. If two questions are naturally asked together, it will ask them together. The order you set only settles ties between questions that are equally worth asking.
Because must-knows are asked first, the questions you mark must-know decide how a conversation opens. Whatever it asks, your bot answers what the lead asked before it asks its next question.
Choice fields need a category group
SELECT and MULTI_SELECT fields draw their options from a category group — a named list
you manage at Business setup → Categories. The same group can serve both sides of a match,
which is what makes overlap-based matching work: a lead picking "wedding" from the same list
your records are tagged from is a comparison the platform can actually make.
Give each option a short description there, so your bot can explain it and suggest the right one to a lead who is not sure: see Describing your options.
Deactivating an option stops it being offered without deleting the history of leads who chose it.
Switching a question off
From the row's menu, Switch off stops your bot asking a question and waiting for it, and keeps every earlier answer. A rule or a matching comparison that uses it pauses too. Switch on brings it back. It works on any question, and is the safe way to drop one of the core questions you do not need. Delete removes it for new leads, though leads keep the answers they already gave. A question cannot be deleted while a rule, a matching comparison or another question's "Only ask when…" depends on it: switch it off, or remove those first.
Changing a field later
Adding, relabelling and reordering are safe and take effect from the next message.
Changing a field's type is the one to be careful with. Existing leads keep whatever they answered, in the old shape. Matching and qualification rules that read the field may not understand the old values afterwards. On a workspace with real leads, adding a new field is usually better than repurposing an old one.
What happens to the answers
They land on the lead as structured data, visible on the lead's page and used by:
- Qualification rules, which decide whether the lead is worth your time.
- Matching dimensions, which compare them against your own records.
Each answer carries a badge when your bot did not simply hear it: Inferred, Suggested, agreed or Confirmed. What those mean is in What your bot finds out.
Amounts are read in the currency your workspace works in (set under your business settings). A lead who writes "50k", "2 lakh" or "€3.000" is read as they meant it. A lead who names a different currency ("50k rupees") is not recorded as if they had said your currency: the bot says plainly that you work in your own currency and asks again, once, and you find what they said on their card as a note ("Budget: 50,000 INR (their currency; this business works in USD)"). If they give it in that currency again, your team is told, the bot stops asking about it and the conversation goes on. Nothing is converted for them. If they had already given a figure and say it was in another currency, the figure is removed.
A question nothing reads is still worth asking if your team reads it — but it will not influence a score.